Data collection, preprocessing and analysis
A workflow of the bioinformatics analysis in this study is shown in Figure 1A. The RNA-sequencing (RNA-seq) and miRNA isoform expression quantification (miRNA-seq) data in the “TCGA-DLBC” project were collected from The Cancer Genome Atlas (TCGA; http://cancergenome.nih.gov/) on December 4, 2018. Raw counts of RNA/miRNA expression data were normalized by trimmed means of M values (TMM) implemented in edgeR and then transformed by voom in limma. Low-expression genes and miRNAs were filtered out; only genes and miRNAs with counts per million reads (cpm) > 1 in more than half of the samples were retained. The RNA-sequencing (RNA-seq) and miRNA isoform expression quantification (miRNA-seq) data of DLBCL contained 48 and 47 DLBCL samples, respectively. After outlier samples and samples with incomplete clinical information were screened out, 46 DLBCL samples remained for subsequent bioinformatics analysis.
Weighted gene coexpression network analysis (WGCNA) was performed in the TCGA-DLBC cohort, and a miRNA-gene interaction network was visualized using Cytoscape v3.4.0. Cox regression and survival analysis were carried out after sample classification according to the mean of miRNA or gene expression level. The RNA-seq data from these samples were subjected to immune cell infiltration profiling using CIBERSORT (16). We used the LM22 leukocyte gene signature matrix, which includes 547 genes distinguishing 22 hematopoietic cell phenotypes and acquired tumor-infiltrating immune cell profiling with a CIBERSORT p value < 0.05.
DLBCL patients enrolled in this study provided informed consent, and specimens were collected at diagnosis biopsy from Shanghai Tongji Hospital Affiliated to Tongji University. None of the subjects received anticancer treatment before biopsy. The protocol was approved by the Institutional Review Board of Center for Medicine, Shanghai Tongji Hospital. All studies were conducted in accordance with the Declaration of Helsinki. Peripheral blood mononuclear cells (PBMCs) were isolated from heparinized whole blood by Ficoll/Hypaque density gradient centrifugation (Solarbio, China) followed by CD8+ T-cell-positive selection using CD8 MicroBeads (Miltenyi, Germany).
The human DLBCL cell lines (LY-1, LY-7) were obtained from the Cell Bank of the Chinese Academy of Sciences (China). The murine B lymphoma cell line A20 was purchased from American Type Culture Collection (ATCC) (USA). LY-1 and LY-7 cells were cultured in Iscove's modified Dulbecco's medium (IMDM, Gibco, USA), and A20 cells were cultured in RPMI 1640 medium (Gibco, USA) supplemented with 10% fetal bovine serum (FBS) (HyClone, USA) and 1% penicillin/streptomycin (HyClone, USA) in a humidified atmosphere of 5% CO2 at 37°C. For LY-7 and A20 cells, 0.05 mM β-mercaptoethanol was added to the culture medium. Primary CD8+ T cells were cultured in RPMI 1640 medium supplemented with 10% FBS, 1% L-glutamine, 1% penicillin/streptomycin and 200 IU/mL IL-2. To stimulate CD8+ T cells, 2 μg/mL of the CMV peptide pool was used for the stimulation of 250,000 cells per well. In direct coculture, CD8+ T cells were harvested and dispensed into 96-well plates according to various effector:target ratios, which were described in the corresponding experiments. LY-1 or LY-7 cells were then added into each CD8+ T cell-containing well at a density of 20,000 cells per well. When the cocultures in ELISA, cytotoxic assay and functional avidity assay were described, CD8+ T cells were preincubated with anti-CD3/anti-CD28 Dynabeads (ThermoFisher, USA) (bead: T-cell ratio = 1:1) overnight and stimulated to achieve substantial expansion. For indirect coculture, tumor cells were seeded into Transwell chambers with a 0.4 μm aperture membrane and then transferred to a 24-well plate seeded with CD8+ T cells in advance, and the supernatant was collected for designed experiments.
Oligonucleotides for miR-340-5p inhibition and forced expression were purchased from GenePharma (China). The specific siRNA, recombinant plasmids KMT5A-OE, FLAG-CD73, HA-COP1, 6x-His-Ub, pLVX-shKMT5A-PURO, pLVX-shCOP1-PURO and their corresponding negative controls were generated and purchased from KeLei Biological Technology (China). The lentivirus was packaged with Δ89 and VSVG helper plasmids, and DLBCL cells were transfected with polybrene, followed by centrifugation at 2,500 × g for 90 min at 37°C. Oligonucleotides, siRNA and plasmids were transfected using Lipofectamine 3000 (Invitrogen, USA) following the manufacturer’s protocols. Cells were subjected to experiments after 24 h of infection. The sequences of shRNA, miRNA mimics and miRNA inhibitors are available in the Supplemental Information (Tables 1-2).
Total RNA was extracted using TRIzol reagent (Invitrogen, USA) by phenol–chloroform precipitation. MiRNAs were reverse transcribed individually by using miRNA-specific reverse transcription primers and the One Step miRNA cDNA Synthesis Kit (HaiGene Bio Inc., China), while total RNA was reverse transcribed into cDNA using the PrimeScript RT Reagent Kit with gDNA Eraser (Takara, Japan). Real-time quantitative RT-PCR was conducted using SYBR Green technology (Takara, Japan) and ABI QuantStudio 6 (USA). U6 and GAPDH were used as endogenous controls for PCR analysis of miRNAs and mRNAs, respectively. Each experiment was run in triplicate. Data were analyzed according to the 2-ΔΔCt method.
Cells were rinsed 3 times with precooled phosphate-buffered saline (PBS) and lysed by RIPA and phenylmethylsulfonyl fluoride (PMSF). Total protein was harvested in 1× sodium dodecyl sulfate (SDS) loading buffer after centrifugation and denaturation. Protein samples were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and electrotransferred to polyvinylidene fluoride membranes (Millipore, USA) with an electrophoresis system (Bio-Rad, CA). The membranes were incubated with horseradish peroxidase (HRP)-conjugated anti-rabbit or anti-mouse secondary antibodies for 1 h at room temperature followed by immunoreactive band detection and analysis.
IHC and ISH
KMT5A and CD8 expression was determined by immunohistochemistry (IHC) for all 40 DLBCL samples. Formalin-fixed paraffin-embedded (FFPE) tissue sections were deparaffinized and dehydrated in xylene and graded ethanol solutions. After deparaffinization, antigen recovery was performed in an autoclave using citrate buffer (pH 6.0) for 15 min, and the slides were then cooled at room temperature and washed in PBS. Slides were incubated with 3% H2O2 and goat serum in the proper order as described in the kit manual (Elabscience, China). Primary antibodies were incubated overnight at 4°C. 3,3’-Diaminobenzidine (DAB) and hematoxylin staining were performed the next day. ISH was performed to detect miR-340-5p expression using an ISH kit (BOSTER, China). Experimental procedures followed the manufacturer’s instructions as previously described (17). Briefly, FFPE samples were stained with DAB and hematoxylin after dehydration and sealing. Oligo (5’ Digoxin-AATCAGTCTCATTGCTTTATAA-3’) was used as the miR-340-5p ISH probe.
For quantification in human specimens and murine models, at least two investigators trained in lymphoma blindly assessed the immunohistochemical staining and achieved a final consensus. Slides were first scanned at low magnification (10x magnification), and 10 high magnification fields (400x magnification) were assessed. For scoring gene and miRNA expression, the intensity of staining was classified into 0 (no expression), 1 (weak expression), 2 (moderate expression) and 3 (high expression), and the percentage of positive cells was categorized as 1 (positive cells ≤25%), 2 (25%< positive cells ≤50%), 3 (50% < positive cells ≤75%) and 4 (positive tumor cells >75%) (18, 19). The histochemical score (H-score) was achieved by multiplying the staining intensity and the percentage of positive cells and ranged from 0 to 300. The mean of H-score was considered the cut-off point (18). Any cell with CD8-positive staining was counted as CD8+ T-TIL. For DLBCL patient samples, the intratumoral area was selected for CD8+ T-TIL evaluation. CD8+ T-TILs were counted manually in each high-power field and scored as follows: 0 (none), 1 (1-2 CD8+ T-TILs), 2 (3-19 CD8+ T-TILs), and 3 (≥20 CD8+ T-TILs) (20, 21).
Antibodies and reagents
The following primary antibodies were used: anti-tubulin (Abcam, ab210797), anti-KMT5A (Abcam, ab111691), anti-CD8 (Abcam, ab17147), anti-CD8 (Abcam, ab217344), anti-CD3 (Abcam, ab16669), anti-ubiquitin (CST, #43124), anti-CD69 (Abcam, ab54217), anti-FLAG (Abcam, ab205606), anti-HA (Abcam, ab236632), anti-6X His (Abcam, ab213204), anti-CD73 (Abcam, ab54217), anti-CD73 (Abcam, ab54217), anti-COP1 (Abcam, ab56400), anti-MKRN1 (Abcam, ab72054), anti-MDM2 (BOSTER, BA3612-2), and anti-Ki-67 (CST, #12202), anti-CD69 (BOSTER, A00529-2), anti-IFN-γ (Abcam, ab231036), anti-IL-2 (Abcam, ab92381), anti-TNF-α (Abcam, ab270264), anti-perforin (Abcam, ab47225), anti-Granzyme B (BOSTER, A00353), anti-perforin (Abcam, ab16074). The following secondary antibodies were used: goat anti-mouse (CST) and goat anti-rabbit (CST). MG132 and cycloheximide (CHX) were purchased from CST (USA). The CMV peptide pool stimulating CD8+ T cells was obtained from Mabtech (Sweden).
Luciferase reporter assay for the 3ʹ UTR study
Luciferase reporter plasmids carrying the wild-type (wt) or mutated (mut) KMT5A 3ʹ UTR were constructed and purchased from Hanbio (China). Two predicted miR-340-5p-binding sites were simultaneously mutated and linked into plasmid carriers. The reporter plasmid was transfected into DLBCL cells along with miR-340-5p mimics or its negative control using Lipofectamine 3000 (Invitrogen, USA). Cells were lysed 48 h after transfection, and luciferase activity was measured using the Dual-Luciferase Reporter Assay System (Promega, USA). The sequences of KMT5A-3ʹ-UTR-wt and KMT5A-3ʹ-UTR-mut are shown in Figure 2E.
CD8+ T cells were cocultured with LY-1 or LY-7 cells (effector:target ratio, E:T = 30:1), and the cell culture supernatant was centrifuged for 20 min at 1,000×g at 2-8°C. The supernatant was collected to carry out the assay. The supernatant was diluted according to preliminary experiments and the detection range for IFN-γ, TNF-α, IL-2. Working solutions of the standard or samples were added to plates and prepared according to the manufacturer’s instructions. The optical density (OD) value of each well was determined simultaneously with a microplate reader set to 450 nm. All of the reagents were obtained from Elabscience (China).
The cytotoxicity assays were conducted using the CytoTox 96 NonRadioactive Cytotoxicity Assay Kit (Promega, USA). According to the protocol, CD8+ T cells were cocultured with DLBCL cell lines LY-1 or LY-7 in a 96-well U-bottom plate at various E:T ratios of 3:1, 10:1, and 30:1 for 4 h. Subsequently, 50 μl of supernatant per well was collected to detect lactate dehydrogenase (LDH) release in a microplate imaging system at an absorbance of 490 nm. As controls, the spontaneous release of LDH was evaluated by incubating T cells or target cells alone, and the maximum release of LDH was assessed by incubating target cells in 0.1% Triton X-100. The results of specific LDH release were calculated as follows: percent specific release = [(experimental OD − effector spontaneous OD − target spontaneous OD)/(target maximum OD − target spontaneous OD)] ×100%.
Functional avidity assessment
The functional avidity of antigen-specific stimulated CD8+ T cells was assessed by limited peptide dilutions and IFN-γ production. The concentration that gives a half-maximal response (EC50) of the peptide to mobilize half of the maximal CD8+ T-cell response was used as the measurement of antigen sensitivity, which was independent of the magnitude of the CD8+ T-cell response with saturated antigen (22). The EC50 of the peptide required to achieve a half-maximal IFN-γ response was determined as previously described (23). Peptide stimulation was performed as described in the Cell culture.
Cells were washed with cold PBS and suspended in PBS containing 10% FBS and 1% sodium azide at a concentration of 1 x 106 cells/mL. Cell staining was performed with antibodies for 30 min, followed by cell washing and flow cytometry analysis. For apoptosis analysis, apoptotic cells were detected using the Annexin V-fluorescein isothiocyanate (FITC)/propidium iodide (PI) Kit (KeyGEN BioTECH, China) according to the manufacturer’s instructions. Briefly, cells with the indicated treatment or transfection were collected and stained with FITC and PI in the dark for 15 min and subjected to flow cytometry. Samples were analyzed using FACSVerse (BD Biosciences) and FlowJo software (Tree Star, USA).
Cells transfected with the indicated lentivirus or oligonucleotides were seeded in 96-well plates for cell viability analysis using Cell Counting Kit-8 (CCK-8; Dojindo, Japan). CCK-8 assays were performed with six replicates, and OD values at 450 nm were measured using a microplate imaging system.
Coimmunoprecipitation and ubiquitination analysis
For endogenous ubiquitination detection, ubiquitination analysis was performed as previously described (24). Briefly, cells were harvested after the indicated treatments and lysed with modified RIPA buffer. After sonication, the lysates were boiled, diluted and centrifuged. The supernatant was subjected to immunoprecipitation with specific antibodies. CD73 ubiquitination and protein detection were determined by western blotting. For the nickel pull-down assay, cell lysates were prepared with lysis buffer. The lysates were sonicated for 30 s, followed by incubation with 50 mL Ni-NTA-agarose (Qiagen, CA) for 4 h at room temperature. The beads were washed with a specific washing buffer, boiled with 2 × SDS loading buffer containing 200 mM imidazole and subjected to western blotting.
Female adult BALB/c mice (4 weeks old, obtained from Shanghai Laboratory Animal Center, Shanghai, China) were injected with 1 × 107 A20 cells into the right flank (12). All mouse experiments were conducted with approval from the Experimental Animal Committee of Shanghai Tongji Hospital. For miR-340-5p, agomirs (GenePharma, China) were delivered on three consecutive days, and intratumoral injections of agomirs and their controls were injected at doses of 30 mg/kg per injection (25). α,β-Methylene adenosine-5′-diphosphate (APCP, Sigma-Aldrich, USA) was injected intratumorally at a daily dose of 20 mg/kg for 1 week followed by twice weekly (26, 27). Treatments started one week after the tumor challenge, and volume measurements started after the tumor reached approximately 0.5 cm × 0.5 cm on the surface (Day 0). Tumor volumes were calculated as 0.5 × a (length) × b (width)2. For flow cytometry, tumor tissues were dissected into approximately 2 mm3 fragments, plated in 24-well plate wells individually and digested using an enzyme mix including DNAse, collagenase, and hyaluronidase. For IHC, tumor tissues were dissected into FFPE or formalin/paraformaldehyde (PFA)-fixed paraffin-embedded sections and subjected to IHC staining and scoring, as described in Immunohistochemistry. Treatments did not cause a significant reduction in body weight (Supplemental Fig. 4).
Statistical analysis was performed using GraphPad Prism software. All data are presented as the mean ± SD of at least three independent experiments. We evaluated the data with Student’s t test. Differences between nonparametric data were analyzed by the Mann-Whitney U test, and multigroup comparisons were performed using one-way analysis of variance (ANOVA). A p value < 0.05 was considered significant. In all charts, the mean and standard error are presented.